Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021)

Original Article

Geo-Spatial Distribution of Frequencies of MTB/RIF Detected Specimens based on Requesting Health Facilities in Manicaland for 2017 and 2018

K Zvinoera 1, J Mutsvangwa2, E Chikaka1, T D Coutinho 3, V Kampira 4, S Mharakurwa1

1. Department of Health Sciences, College of Health Agriculture and Natural Sciences Africa University, Zimbabwe 2. Biomedical Research and Training Institute, Zimbabwe 3. Department of Geography and Environmental Science University of Zimbabwe 4. Provincial Hospital, Ministry of Health Zimbabwe

ABSTRACT Genexpert CPU crushed. Geographical Positioning System (GPS) of the health facilities were Objectives: The aim of this study was to produce recorded.The study used MTB detected frequencies Geo Spatial Distribution of Frequencies of at a facility in relation to surrounding facilities in MTB/RIF Detected Specimens based on Requesting Manicaland, then ran optimised hotspot analysis Health Facilities in Manicaland Zimbabwe for 2017 function in Arc Map 10.5 to implement the Gi* and 2018, so as to give insight to TB program statistic. managers. Focusing elimination interventions on hot pockets of Tuberculosis (TB) strengthens Results: Overall provincial MTB detected rationale use of resources in resource limited positivity was 2221/36055 (6.2%).Overall countries like Zimbabwe. Early detection and early provincial Rifampicin Resistant (RR) positivity was treatment is backbone of breaking TB transmission. .111.2221(5.0%).Geo-spatial map of Manicaland Drug resistant tuberculosis (DRTB) control showed 10 facilities that are RR hotspots with 7/10 interventions like Programmatic Management of (70%) of the facilities in district. Drug Resistant TB or mentoring on Short, all Oral district had facilities that were MTB detected high Regimen for Rifampicin resistant Tuberculosis hotspots.For the whole of Manicaland, Buhera (ShORRT) will be driven by science. district had100% MTB detected low hotspots facilities. Ninety percent hotspots were clustered Materials and Methods: The retrospective study around 2 of the 15 Genexpert Sites in Manicaland, was carried out in Manicaland, Zimbabwe. namely Mission Hospital and Manicaland one of the 10 provinces in Zimbabwe, Hospital. has 7 districts with 308 health facilities. During this retrospective cross sectional study 2221 MTB Conclusion: Study identified health facilities with detected results of 2017 and 2018, downloaded from high frequencies of RR areas. For the identified 14 of the 15 Genexpert sites in Manicaland were health facilities with high frequencies of RR employed to generate hotspot maps. Fifteenth specimens, NTP may focus DRTB control Genexpert site lost its electronic records when interventions like PMDT, or mentoring on ShORRT. For the health facilities with high frequencies of Corresponding author: MTB detected NTP can focus trainings in TB Case Zvinoera Katherine Cell phone +263777398293 Key Words: Genexpert MTB/Rif, Geographical Positioning Email- [email protected] System, Geo Spatial Mapping

78 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021)

Management. Instead of uniformly spreading the rationale prepositioning of resources, a must in limited resources to all 325 facilities, efforts resource limited countries like Zimbabwe. streamlined to manageable number of 20 facilities in Resources like TB case management training, commensurate with identified gap( e.g. objective Programmatic Multi Drug Resistant Tuberculosis selection of cadres for training, data driven training, Mentoring through on site supportive supportive supervision & targeted awareness visits, community awareness campaigns in targeted campaigns). communities7. Geo Spatial Distribution of Frequencies of MTB/RIF Detected Specimens INTRODUCTION based on Requesting Health Facilities in Utilization of geo-spatial mapping enables Manicaland had not been created, judging by the assessing of geographical distribution of infectious lack of literature. There was need to generate geo- diseases. Tuberculosis (TB)like other infectious spatial distribution mapping of Genexpert diseases, have epidemiological patterns that differ MTB/Rif's semi quantitative results in Manicaland. across geographical regions.1,2,3Early TB detection Question on hand was what the Geo Spatial and early treatment is the backbone to breaking Distribution of Frequencies of MTB is/RIF transmission. In attempts to manage TB, the World Detected Specimens based on Requesting Health Health Organization (W.H.O.)'s End TB Strategy, Facilities in Manicaland Zimbabwe for 2017 and has a goal to end the TB epidemic by reduction to 2018? The aim of this study was to generate geo- <10 cases per 100 000 population by 2035.4 spatial distribution maps of Genexpert MTB/Rif Zimbabwe is among the 14 countries with a triple results in Manicaland, Zimbabwe so as to give burden of TB, Human Immunodeficiency Virus insight to TB program managers. In the study a positive people co-infected with TB (TB/HIV) and spatial map was generated showing the Geo Spatial multi-drug resistant TB (MDR-TB). Incidence of Distribution of Frequencies of MTB/RIF TB in Zimbabwe in 2018 was estimated to be 210 Detected Specimens based on Requesting Health per 100,000 population. Further estimates were that Facilities in Manicaland Zimbabwe for 2017 and 62% of patients diagnosed TB, were HIV co- 2018. Focusing elimination interventions on hot infected. Lastly among multi drug resistant TB pockets of Tuberculosis (TB) strengthens rationale (MDR-TB) patients HIV prevalence was 80%5. use of resources in resource limited countries like Zimbabwe. Early detection and early treatment is To end TB successfully, disease elimination backbone of breaking TB transmission. Drug campaigns need to be characterized by locally resistant tuberculosis (DRTB) control interventions tailored responses that are informed by evidence like Programmatic Management of Drug Resistant based medicine(e.g. objective selection of cadres for TB or mentoring on Short, all Oral Regimen for training, data driven supportive supervision & Rifampicin resistant Tuberculosis (ShORRT) will targeted awareness campaigns).For such a response be driven by science. to tuberculosis, a three-step process is vital, which includes, using existing programmatic data to guide MATERIALS AND METHODS decisions, collection of additional data (e.g. geographic information, drug resistance, and risk The study design employed was retrospective cross factors) to aid creation of tailored responses. Geo- sectional design. The study was carried out in Spatial distribution maps improve understanding of (figure1). Figure 1 has TB transmission dynamics. Focus on eliminating Zimbabwe map on top left corner, with Manicaland hot spots of TB is one strategy to meet the WHO end located on the Eastern part of Zimbabwe. Figure 1 TB target of zero death due to TB by 20356. has enlarged map of Manicaland, which covers a Decisions made from spatial mapping include total area of 36,459 square kilometres. It is the

79 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021) second most populous province after Harare with a MTB detected very low, MTB detected trace. Each population total of 1.75million (Census, 2012). It is followed by rifampicin resistant result (RR); RR not the third most densely populated province after detected RR detected or RR indeterminate. Harare and . According to Zimbabwe District Health Information System (DHIS2) the The geographic locations of health facilities were province has 308 health facilities in 7 districts collected using Google earth pro. The geographic served by 15 Genexpert sites. The study population coordinates for Manicaland health facilities were was MTB/Rif results analysed in Manicaland for the then entered in a spread sheet and converted into a period 2017 and 2018. Complete enumeration was comma-separated values (csv) file for use in a GIS used. Study data sources were Laboratory TB environment. The points were mapped using arcGis registers, Genexpert machines and GPS co- version10.5 for visual analysis of MTB cases. The ordinates captured for the 308 requesting health study used MTB detected frequencies at a facility in facility. Unit used for recording coordinates was relation to surrounding facilities to run optimised requesting facility. hotspot analysis function in Arc Map 10.5 to implement the Gi* statistic. The Getis-Ord (Gi*) statistic assesses the extent to which events such as MTB data exhibit identifiable spatial patterns in space as hotspots and cold spots. Hotspots occur when areas with high values are surrounded by high MTB values while cold spots are locations with typically low values surrounded by similarly low values, from a given location i at spatially varying distances. The optimised hotspot analysis function

in Arc Map 10.5 was used to implement the Gi* statistic. The selection of the optimised hotspot analysis was based on its ability to correct for multiple testing as well as spatial dependence. The Z-scores and p-values measure statistical significance, influencing the decision whether to reject or fail to reject the null hypothesis. A high z- score and small p-value for a feature indicates a spatial clustering of high values. A low negative z- Figure1: Districts of Manicaland Province. score and small p-value indicates a spatial clustering of low values. The higher the z-score, the more intense the clustering. A z score near zero indicates 8 Genexpert MTB/Rif Excel sheet data was no apparent spatial clustering .The study was downloaded directly from the Genexpert machines approved by Manicaland Provincial Directorate, to electronic format. Data from paper based TB Africa University Research Ethics Committee registers was merged into the electronic format, the (AUREC) and Medical Research Council of excel sheet and de-identified. Data was entered as Zimbabwe (MRCZ). line item capturing independent variables and RESULTS outcome or depended variables. MTB/Rif data produced a categorical variable of; MTB detected Figure 2 shows how we downloaded 48260 high,MTB detected medium, MTB detected low and Genexpert MTB/Rif results and ended up with 2221

80 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021) results for this study. The findings in this study were Health Facilities that Submitted High overall provincial MTB detected positivity was frequencies of Rifampicin Resistant (RR) 2221/36055 (6.2%).Overall provincial Rifampicin Specimens in Manicaland for 2017 and 2018 Resistant (RR) positivity was 111/2221(5.0%). The red exes are the areas of concern in Figure 3. Our findings suggest that 7 facilities in , 2 in Chipinge and 1 in ; D ownloaded Munyanyi clinic, Rambanepasi clinic, 48260 from 14 Gene Mombeyarara Rural Health Centre, Gombe clinic, Xpert Sites Murambinda Mission Hospital, Nyashanu Mission Clinic, Chiwese clinic, Matotwe Rural Health Centre, Kopera Rural Health Centre, Gaza clinic. Removed unsuccessful There was intra-provincial epidemiological results like error, invalid difference according to geographical setting. Intra- and no results provincial epidemiological difference of RR allows focusing of MOHCC interventions to problem areas.NTP needs to further interrogate these findings to rule out practices leading to manmade

drug resistance in the facilities that are RR hot spots. 36055 Other covariates have to be investigated such as

facility ownership and specimen transport network

at facilities.

Health Facilities that Submitted High frequencies of MTB Detected High Specimens in Manicaland for 2017 and 2018 Removed MTB not detected Semi quantitative MTB results gave further insight; the GPS mapping data showed which requesting

health facilities were surrounded by higher or lower concentrations of TB. One of the indicators of how 2221 good the TB program is at a health facility is reflected by diagnosis of TB at the earliest stages of the disease when the semi quantitative result is still trace or low, instead of high or moderate. A health facility which diagnoses mostly MTB detected high or MTB detected medium is an indication of poor performance of the TB program at the health facility. Facilities identified as TB hot spots require interventions to break chain of transmission to be applied or invested to decrease chances of community exposure to TB. May be sign of low Geo-Spatial index of TB suspicion at facility or may be poor Mapping health seeking behaviour of population near MTB detected high hot spots. Figure 2: Flow Chart of Study Samples

81 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021)

Our findings in figure 4, suggest that Chipinge Health Facilities that Submitted High district has 9 of the 14(64.3%) health facilities frequencies of MTB Detected Low Specimens in identified as MTB detected high hot spots in Manicaland for 2017 and 2018 Manicaland.These results are not comparable to any Our findings in figure 5, suggest that Buhera has in Zimbabwe, due to lack of literature on similar hotspots of MTB detected low at Chapwanya studies. Chirenda et al 2020 also carried out spatial clinic,Garamwera clinic, Murambinda Mission mapping, but of TB patients archived in the Harare Hospital,Gombe clinic, Nyashanu Mission Clinic, City registry. Their findings were that the home Mombeyarara Rural Health Centre and Munyanyi addresses of the TB patients formed hotspots in the clinic.These findings may be sign of good TB peri urban area where there were no amenities. The program at health facility with high index of lowest unit for our study was requesting health suspicion.Another plausible reason may be early facility. The findings show that the hotspots are health seeking behaviour of catchment facilities that form a cluster around the Chipinge population.MTB detected low incident showed that district hospital Genexpert site. The plausible reason TB disease was diagnosed before the patient had a could be that patients move (patient referral chance to infect others in the community. followed by specimen referral to Genexpert site) to clinics served by reliable integrated specimen Health Facilities that Submitted High transport system (IST). That means patients could frequencies of MTB Detected Very Low be transient and not permanent residence of Specimens in Manicaland for 2017 and 2018 requesting facilities. High bacillary density facilities could be due to low index of suspicion. Or Our findings in figure 6, suggest that Districts could be low health seeking behavior of the represented as follows; 3 requesting facilities from population around the facility. Further research has Chipinge district, 7 requesting facilities from to be carried out to get more insight, however these Buhera district, 2 requesting facilities from Mutasa findings have allowed stream lining of elimination district. intervention efforts. The intervention efforts could MTB detected low incident shows TB disease be training clinicians manning the hotspots to TB diagnosed before the patient transmits TB in Case Management, mentoring via telemedicine or community. The results of MTB detected very low supportive visits. Hosting awareness campaigns are not conforming to the expectations set by the such as the World TB day in hotspots would be other hotspot maps in this study. The plausible another intervention strategy.These initial results reason for these unexpected results, could be help NTP to narrow down on a few health facilities, supporting the fact that it's not catchment which can be further investigated. Consolidated population, but transient patient who temporarily MTB results hot spots further investigation should move to near Genexpert sites to seek Genexpert establish if specimens were submitted from local MTB/Rif testing services. population, or could be people flocking from outside the district or the province. Rule out movement of CONCLUSION people from within the district for temporary Our findings demonstrate that there is variation in residence at convenient health facilities served by epidemiological pattern of TB in Manicaland. NTP integrated specimen transport. NTB has fuel may mobilise resources to enable focused TB coupons to supplement motorized EHTs. The elimination interventions in the identified facilities identified hot spot facilities could have population at risk such as prison or miners or people dwelling in that are hot spots. NTP be guided by science to overcrowded conditions. enable rational use of its limited resources. The

82 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021) package of intervention may be more streamlined or distribution of people diagnosed with specific type of elimination intervention strategy tuberculosis through routine and active case incommensurate with identified gap i.e. hotspot for finding: a community-based study in Kampala, RR or MTB detected high or MTB detected low. Uganda. Infect Dis Poverty. 2020 Jun 22; 9(1):73. doi: 10.1186/s40249-020-00687-2. RECOMMENDATIONS PMID: 32571435; PMCID: PMC7310105. Based on these findings M.O.H.C.C. to make use of 3. Maciel EL, Pan W, Dietze R, Peres RL, Vinhas the evidence generated to distribute the scarce TB SA, Ribeiro FK, Palaci M, Rodrigues RR, elimination intervention resources to the identified Zandonade E, Golub JE. Spatial patterns of hot spots (e.g. objective selection of cadres for pulmonary tuberculosis incidence and their training, data driven supportive supervision & relationship to socio-economic status in targeted awareness campaigns). NTP be guided by Vitoria, Brazil. Int J Tuberc Lung Dis. 2010 science to enable rational use of its limited Nov; 14(11):1395-402. PMID: 20937178; resources. The package of intervention may be more PMCID: PMC3713790. streamlined or specific type of elimination 4. Zimbabwe Tuberculosis National Strategic intervention strategy in commensurate with Plan (2017 to 2020) Ministry of Health and identified gap i.e. hotspot for RR or MTB detected Child Care high or MTB detected low. For the identified health 5. Timire C, Sandy C, Kumar AMV, Ngwenya M, facilities with high frequencies of RR specimens, Murwira B, Takarinda KC, Harries AD. Access NTP may focus DRTB control interventions like to second-line drug susceptibility testing PMDT, or mentoring on ShORRT. For the health results among patients with Rifampicin facilities with high frequencies of MTB detected resistant tuberculosis after introduction of the ® NTP can focus trainings in TB Case Management. Hain Line Probe Assay in Southern provinces, Zimbabwe. Int J Infect Dis. 2019 Apr; 81:236- Future Research 243. doi: 10.1016/j.ijid.2019.02.007. Epub Suggestion for future research is to carry out a 2019 Feb 15. PMID: 30776546. research focusing on hotspots to enable covariate 6. Theron G, Jenkins HE, Cobelens F, Abubakar I, analysis. Khan AJ, Cohen T, Dowdy DW. Data for action: collection and use of local data to end Acknowledgements tuberculosis. Lancet. 2015 Dec 5; Biomedical Research and Training Institute (BRTI) 386(10010):2324-33. doi: 10.1016/S0140- for facilitating the research 6736(15)00321-9. Epub 2015 Oct 26. PMID: This study was funded by Trials of Excellence in 26515676; PMCID: PMC4708262. Southern Africa II (TESA II). 7. Chirenda J, Gwitira I, Warren RM, Sampson SL, Murwira A, Masimirembwa C, Mateveke REFERENCES KM, Duri C, Chonzi P, Rusakaniko S, et al. 1. Global tuberculosis report 2018. World Health Spatial distribution of Mycobacterium Organization 2018. Available from: Tuberculosis in metropolitan Harare, https://apps.who.int/iris/handle/10665/274453 Zimbabwe. PLoS One. 2020 Apr 21; 2. Robsky KO, Kitonsa PJ, Mukiibi J, Nakasolya 15(4):e0231637. doi: 10.1371/journal. O, Isooba D, Nalutaaya A, Salvatore PP, pone.0231637. PMID: 32315335; PMCID: Kendall EA, Katamba A, Dowdy D. Spatial PMC7173793.

83 Medical Journal of Zambia, Vol. 48 (2): 78 - 84 (2021)

8. Getis, A. & Ord, J.K. (1992). The analysis of ED. Identifying locations of recent TB spatial association by use of distance t r a n s m i s s i o n i n r u r a l U g a n d a : a statistics, Geographical Analysis, 24, 189- multidisciplinary approach. Trop Med Int 206.http://dx.doi.org/10.1111/j.1538- Health. 2015 Apr; 20(4):537-45. doi: 4632.1992.tb00261.x 10.1111/tmi.12459. Epub 2015 Feb 4. PMID: 9. Chamie G, Wandera B, Marquez C, Kato- 25583212; PMCID: PMC4355181. Maeda M, Kamya MR, Havlir DV, Charlebois

84